Functional variants of DOG1 control seed chillingresponses and variation in seasonal life-historystrategies in Arabidopsis thalianaAlejandra Martínez-Berdejaa, Michelle C. Stitzera,b, Mark A. Taylora, Miki Okadaa, Exequiel Ezcurrac,Daniel E. Runcied, and Johanna Schmitta,b,1

aDepartment of Evolution and Ecology, University of California, Davis, CA 95616; bCenter for Population Biology, University of California, Davis, CA 95616;cDepartment of Botany and Plant Sciences, University of California, Riverside, CA 92521; and dDepartment of Plant Sciences, University of California, Davis,CA 95616

Contributed by Johanna Schmitt, November 22, 2019 (sent for review July 19, 2019; reviewed by Leonie Bentsink, Xavier Picó, and Peter Tiffin)

The seasonal timing of seed germination determines a plant’s re-alized environmental niche, and is important for adaptation toclimate. The timing of seasonal germination depends on patternsof seed dormancy release or induction by cold and interacts withflowering-time variation to construct different seasonal life histo-ries. To characterize the genetic basis and climatic associations ofnatural variation in seed chilling responses and associated life-historysyndromes, we selected 559 fully sequenced accessions of the modelannual species Arabidopsis thaliana from across a wide climate rangeand scored each for seed germination across a range of 13 cold strat-ification treatments, as well as the timing of flowering and senes-cence. Germination strategies varied continuously along 2 majoraxes: 1) Overall germination fraction and 2) induction vs. release ofdormancy by cold. Natural variation in seed responses to chilling wascorrelated with flowering time and senescence to create a range ofseasonal life-history syndromes. Genome-wide association identifiedseveral loci associated with natural variation in seed chilling re-sponses, including a known functional polymorphism in the self-binding domain of the candidate gene DOG1. A phylogeny ofDOG1 haplotypes revealed ancient divergence of these functionalvariants associated with periods of Pleistocene climate change, andGradient Forest analysis showed that allele turnover of candidateSNPswas significantly associatedwith climate gradients. These resultsprovide evidence that A. thaliana’s germination niche and correlatedlife-history syndromes are shaped by past climate cycles, as well aslocal adaptation to contemporary climate.

delay of germination | germination niche | seed dormancy | genome-wideassociation | stratification

The seasonal timing of seed germination is a niche construc-tion trait that shapes phenology and life history throughout

the life cycle (1–10), and may be critical for adaptation to climate(11–13). Germination timing is under strong selection in nature(1, 5, 13–17) and determines the selective environment for sub-sequent life-history traits, such as flowering time (1, 15, 16, 18),driving correlated evolution of suites of life-history traits acrossclimate gradients (11, 12, 19, 20). Seasonal germination timing inthe wild depends upon natural variation in primary dormancy atdispersal and in responses to environmental cues that mediateannual cycles of dormancy release and secondary dormancy in-duction. In particular, variation in seed responses to seasonalchilling cues can determine whether seeds germinate in fall orspring (21–24). Fall vs. spring germination defines distinct life-history strategies whose adaptive value may vary across climates(11, 12). To understand the role of these critical niche con-struction traits in climate adaptation, it is important to describetheir genetic basis. Here we characterize natural variation inseed chilling responses and correlated life-history traits in theannual species Arabidopsis thaliana, identify genomic variantsunderlying these traits, and test whether these variants exhibit asignature of adaptation to climate across the species range.

In annual plants, variation in seasonal germination timingcreates alternative life-history strategies (SI Appendix, Fig. S1).Winter annuals germinate in autumn, overwinter, and then flowerand disperse seed in spring, whereas spring or summer annualsoverwinter as seeds and germinate, flower, and disperse seed inspring or summer (11, 17, 21, 23). Mixtures of fall and springgermination cohorts are also observed within populations (7, 10,11, 17, 21), a form of within-year bet hedging (25). Whether andwhen seeds germinate in a given seasonal environment dependsupon the germination niche, the range of conditions under whichgermination is possible (7, 11, 21, 26). Primary dormancy of freshlydispersed seeds varies among genotypes and seed maturationenvironments (22, 23, 27–33). Release from primary dormancymay occur through after-ripening at warm ambient temperatures,or through short exposure to chilling (21, 22, 34, 35). Both of thesemechanisms allow germination of winter annual seedlings in fall.In many winter annuals, seeds remaining in the soil seed bank infall enter secondary dormancy when exposed to prolonged winterchilling, preventing winter and spring germination (21, 22, 26). In

Significance

The seasonal timing of seed germination is critical for plantfitness in different climates. To germinate at the right time ofyear, seeds respond to seasonal environmental cues, such ascold temperatures. We characterized genetic variation in seeddormancy responses to cold across the geographic range of awidespread annual plant. Induction of secondary seed dor-mancy during winter conditions (which restricts germination toautumn) was positively correlated with flowering time, con-structing winter and spring seasonal life-history strategies.Variation in seed chilling responses was strongly associatedwith functional variants of a known dormancy gene. Thesevariants showed evidence of ancient diversification associatedwith Pleistocene glacial cycles, and were associated with cli-mate gradients across the species’ geographical range.

Author contributions: A.M.-B. and J.S. designed research; A.M.-B. and M.O. performedresearch; A.M.-B., M.C.S., M.A.T., E.E., D.E.R., and J.S. analyzed data; and A.M.-B., M.C.S.,and J.S. wrote the paper.

Reviewers: L.B., Wageningen University & Research; X.P., Estación Biológica de Doñana(Consejo Superior de Investigaciones Cientifica); and P.T., University of Minnesota.

The authors declare no competing interest.

This open access article is distributed under Creative Commons Attribution-NonCommercial-NoDerivatives License 4.0 (CC BY-NC-ND).

Data deposition: Germination, phenology and DOG1 haplotype data are available onDryad repository, https://doi.org/10.25338/B8VS4P. DOG1 haplotype data and code arealso available on GitHub, https://github.com/mcstitzer/martinez-berdeja_dog1.1To whom correspondence may be addressed. Email: jschmitt@ucdavis.edu.

This article contains supporting information online at https://www.pnas.org/lookup/suppl/doi:10.1073/pnas.1912451117/-/DCSupplemental.

www.pnas.org/cgi/doi/10.1073/pnas.1912451117 PNAS Latest Articles | 1 of 9

EVOLU

TION

Dow

nloa

ded

by g

uest

on

May

19,

202

0

contrast, summer annual seeds lose dormancy with chilling overwinter, allowing germination in spring (21, 23). Seeds that losedormancy in fall and do not enter secondary dormancy over thewinter may germinate in either fall or spring resulting in a mixtureof fall and spring annual life histories. Natural variation in dor-mancy responses to seasonal chilling cues thus determines thegermination niche and the expression of seasonal life-history strat-egies within and among populations in different environments.Germination timing also determines the selective environment

for subsequent life-history traits, such as flowering time (1, 15,16, 18). Seedlings that germinate in fall may be selected to delayflowering until winter is past, which may favor late flowering andstrong vernalization (winter chilling) requirements. However,strong vernalization requirements and late flowering may bemaladaptive for summer annuals, especially if the growing sea-son is short. Consequently, natural selection on germination andlater life-history traits may be correlated, and adaptation to cli-mate across a species’ range may therefore involve coordinatedevolution of suites of life-history traits (11, 12, 19, 20, 36, 37).To understand the adaptive evolution of the germination niche

across different climates, it is important to elucidate its geneticbasis. It is also of interest to examine the genetic basis of correlatedtraits that may form adaptive life-history syndromes, such as wintervs. summer annual life histories. If different traits share commongenetic mechanisms, this pleiotropy may constrain or facilitateadaptive evolution of multivariate syndromes (38). On the otherhand, if the loci contributing to variation in different traits do notoverlap, then adaptive divergence of life-history syndromes couldonly occur through correlated selection response at different sets ofloci across a climate gradient. Once loci underlying life-historyvariation are identified, we can test for a geographic signature ofselection by climate. If allelic variants at those loci are involved inadaptation to climate, we expect them to be significantly associatedwith relevant climate variables across the landscape (39–41).The model plant A. thaliana is an ideal system for dissecting

the genetic and environmental determinants of climate ad-aptation in the germination niche and correlated life-historytraits. Extensive sequence data are available for this annualspecies, facilitating genome-wide association (GWA) (42, 43).A. thaliana inhabits a wide native climate range across Eurasiaand Africa, and has shifted across the landscape with Pleistoceneclimate cycles, repeatedly contracting and expanding out of glacialrefugia (44–47). Across this range, the species exhibits substantiallife-history variation (11, 13, 48). The pathways involved in seeddormancy and flowering time have been characterized throughfunctional studies (24, 49). Natural variation in primary dor-mancy and after-ripening is also well documented (11, 12, 19, 20,50), and underlying allelic variants have been identified (13, 27,50–54). However, less is known about the genetic basis of naturalvariation in seed chilling responses in this species, despite theirimportance for fine-tuning dormancy cycles to shape seasonalgermination timing (19, 22).DELAY OF GERMINATION 1 (DOG1), a key regulator of

seed dormancy and a member of a small gene family of unknownmolecular function (27), is a particularly important candidategene for natural variation in germination niche traits. This genecontrols primary seed dormancy through multiple mechanisms(55–62). Allelic variants of DOG1 are associated with naturalvariation in primary dormancy, after-ripening, and germinationtiming in the field (5, 13, 50, 52, 63). DOG1 expression is asso-ciated with dormancy variation measured as after-ripening time(27, 53) and exhibits clinal variation; southern ecotypes havehigher DOG1 expression associated with longer after-ripening(12, 51, 63). DOG1 variation is also associated with naturalvariation in flowering time (42, 43, 64) and may have pleiotropiceffects both indirectly through cascading effects of germinationtiming (6) and directly through its effects on levels of associatedmicro RNAs (60). However, little is known about the contribution

of DOG1 functional polymorphisms to natural variation in ger-mination responses to chilling.Here we combine experimental phenotypic data from a geo-

graphically diverse set of accessions, whole-genome polymorphismdata (43), and geographic and climate information to address thefollowing questions:

1) What is the range of natural variation of germination re-sponses to chilling? How does natural variation in germina-tion responses to chilling covary with other life-history traitsto shape winter and spring annual life-history syndromesacross the landscape?

2) What is the genetic basis of natural variation in germinationresponses to chilling and associated life-history traits? Doesallelic variation in DOG1 contribute to variation in seed chill-ing responses to shape the seasonal germination niche?

3) Do these genetic variants exhibit a geographic signature ofadaptation to climate?

Results and DiscussionNatural Variation in Seed Chilling Responses Creates Diverse SeasonalLife Histories.We grew 559 Arabidopsis accessions to reproductivematurity at 14 °C with a 12/12-h photoperiod following 6 wk ofvernalization at 4 °C. Fresh seeds were harvested from each plantwhen siliques matured, and were tested for germination at 22 °Cimmediately (i.e., base germination), or after 4, 8, 11, 15, 22, or32 d of dark stratification at 4 °C and 10 °C. We performed aprincipal component analysis (PCA) on all germination pheno-types. The first 2 principal components (PCs) of germinationtiming and dormancy level explained 91% of the germinationvariation (PC1germ = 83%, PC2germ = 7%) (Fig. 1 and SI Ap-pendix, Fig. S2). We also measured days to flowering (DTF),and days to senescence (DTS) of the maternal plants. Addi-tionally, we assayed germination of dry-stored seeds at 6-wk in-tervals to quantify after-ripening requirements (days of seed drystorage required to reach 50% germination, DSDS50) (Table 1).All germination variables measured were positively associated

to PC1germ, (SI Appendix, Table S1), and ecotypes with a positivescore along this axis had overall high germination. Accessions withhigh overall germination (PC1germ) were late-flowering and late-senescing and had short after-ripening times, whereas accessionswith low germination were early-flowering and early-senescingand had long after-ripening times (Table 1). Variation in over-all germination fraction reflects variation in primary dormancy,consistent with previous observations of natural variation indormancy measured as after-ripening requirement (11, 12, 50).Accessions with high germination were distributed in central andnorthern Europe (SI Appendix, Fig. S3A). Low dormancy allowsimmediate germination after dispersal, which is favored by se-lection in northern climates (13) and may result in multiplegenerations per year in midlatitudes with wet summers (7, 65). Incontrast, primary dormancy that cannot be broken by chillingexposure, correlated with strong after-ripening requirements,would maintain a large population in the seed bank (7), a po-tential bet-hedging strategy (66). Accessions with low germina-tion were distributed in south and central Spain and southernEurope (SI Appendix, Fig. S3A).PC2germ represents the germination response to cold. Base

germination and germination percent of all stratification times at10 °C were positively associated with PC2germ, while germinationpercent at 11 to 32 d at 4 °C were negatively associated to thisaxis (SI Appendix, Table S1). Accessions with a positive score onPC2germ had high base germination and secondary dormancyinduced by prolonged cold exposure at 4 °C (Fig. 1 and SI Ap-pendix, Fig. S2 and Table S1), flowered and senesced later, andhad longer after-ripening periods (Fig. 2 and Table 1). Theseaccessions would behave as winter annuals, with germination

2 of 9 | www.pnas.org/cgi/doi/10.1073/pnas.1912451117 Martínez-Berdeja et al.

Dow

nloa

ded

by g

uest

on

May

19,

202

0

restricted to late summer or fall, and flowering and seed dis-persal the following summer. Many of these accessions occur inScandinavia and the Iberian Peninsula (SI Appendix, Fig. S3B).In contrast, accessions with a negative score on PC2germ thatbreak primary dormancy with brief chilling exposure and do notenter secondary dormancy with prolonged chilling also displayedfast phenological transitions and had long after-ripening times,preventing summer germination. Dormancy release by coldwould allow germination in both fall and spring, and coupledwith rapid reproduction, would facilitate rapid cycling or springannual life histories. Seeds that do not germinate in fall wouldoverwinter in the soil under cold temperatures and would beready to germinate the next spring (26). This seasonally flexiblegermination may be favored in disturbed, ruderal landscapes.These accessions were common at midlatitudes in Europe andEngland (SI Appendix, Fig. S3B).The timing of germination defines the temporal environment an

annual plant experiences, and interacts with later life-cycle traitsto generate different life histories. Our results are consistent with

previous observations of covariation in seed dormancy (measuredas after-ripening requirements) and flowering-time traits acrossclimatic gradients in A. thaliana (12, 19, 20, 37). Trait syndromesallow local adaptation, as covariation in life-history traits, growthrate, and stress responses influence fitness (37). However, our ob-servations of germination responses to chilling add a new dimensionto this picture of multivariate life histories, showing that cold-induced secondary dormancy is correlated with flowering and se-nescence to construct a wide range of seasonal life-history strategies.

Genetic Architecture of Seed Responses to Chilling: A Major Role forDOG1. To understand the genetic architecture of the germinationniche, we performed GWA analyses on the germination PCs using498 fully sequenced A. thaliana accessions and 3,483,598 singlenucleotide polymorphisms (SNPs; 1001 Genomes Consortium). Aregion on chromosome (Chr.) 5 was the most highly associated toboth the first and second PCs of germination. This region included1 SNP significantly associated with PC1germ and 42 SNPs signifi-cantly associated with PC2germ. Several of these SNPs likely tag thesame functional variant, as they are organized into only 4 linkagedisequilibrium (LD) blocks spanning multiple candidate genes.Different SNPs were most highly associated to PC1germ andPC2germ, although the closest gene to both was DOG1 (Fig. 3 andSI Appendix, Tables S2 and S3). The SNP significantly associatedto PC1germ and PC2germ (Chr. 5: 18,592,365) (SI Appendix, TableS2) had previously been shown to tag the promoter region ofDOG1 and to be associated with after-ripening time variation (50)(SI Appendix, Fig. S4). However, in our experiments, DSDS50 wasassociated with a different set of SNPs in several LD blocks, whichdid not include DOG1 (SI Appendix, Fig. S5 and Table S2). Usingmore permissive thresholds, we found that the top 1,000 SNPs

A

B

Fig. 1. (A) PC scores of germination variables (base, and cold treatments: 4, 8, 11, 15, 22, and 32 d at 4 and 10 °C). Percent variance explained by each axis:PC1germ = 83%, PC2germ = 7%. (B) Mean percent germination in each treatment for the lines with the 10% highest and lowest scores on PC1germ and PC2germ.Colors indicate the amino acid sequence at the DOG1 self-binding domain (53), which we use to define haplotypes, and grey color indicates accessions forwhich we were not able to assign a haplotype.

Table 1. Correlations between the first 2 PCs from thegermination variables PCA with phenology variables of themother plants

Variable

PC1germ PC2 germ

df r P value r P value

DTF 523 −0.07 0.09 0.43 <2.2 × 10−16

DTS 523 0.18 5.19 × 10−5 0.50 <2.2 × 10−16

DSDS50 517 −0.49 <2.2 × 10−16 −0.43 <2.2 × 10−16

Martínez-Berdeja et al. PNAS Latest Articles | 3 of 9

EVOLU

TION

Dow

nloa

ded

by g

uest

on

May

19,

202

0

associated to each trait were enriched for seed-specific gene on-tologies (SI Appendix, Tables S4 and S5).As SNPs near DOG1 were associated in both analyses, we

further characterized genetic variation at DOG1. Known indelpolymorphism within the self-binding domain of DOG1 (aminoacid positions 13 to 16) (53) is not fully captured by publishedSNP data for A. thaliana (43, 45, 46). We therefore locallyreassembled exon 1 of DOG1 using raw sequence data from the1001 Genomes Consortium (43), as well as African (45) andChinese (46) A. thaliana accessions. This indel appears to befunctional, as it likely affects dimerization of DOG1 proteins andalters their incorporation into larger protein complexes (53).Haplotypes in the self-binding domain have been defined by theamino acid variants present. We identified the ancestral self-binding DOG1 E-haplotype (ECCY) and 2 loss-of-functionDOG1 D-haplotypes (D-RY, D-SY) previously reported (53),as well as 2 additional rare derived haplotypes without an aminoacid deletion in the self-binding domain (ECSY and EFSY)found in less than 5% of individuals and restricted to Iberia (SIAppendix, Figs. S6 and S7). DOG1 haplotype identities are sig-nificantly associated with PC2germ scores (SI Appendix, TableS6A). The loss-of-function D-RY and D-SY DOG1 haplotypesshare the reference allele of 3 of the most highly associated SNPsfor PC2germ in the same LD block (Chr. 5: 18,590,327, 18,590,741,

18,590,743), while the alternative allele is associated to the self-binding E-haplotypes (SI Appendix, Fig. S8 and Table S2). Al-though these SNPs tag the indel and nonsynonymous variant thatdifferentiates D- and E-containing haplotypes, our haplotypegenotyping distinguishes further fine-scale genetic differencesbetween accessions. We expect this is particularly important inrefining the self-binding ability of the DOG1 protein, which isknown to play an important role in the germination response tochilling (53, 59). Accessions carrying the self-binding DOG1 E-haplotypes were characterized by strong secondary dormancyinduced by prolonged chilling (high PC2germ) (Fig. 2); in con-trast, the loss-of-function DOG1 D-haplotypes, had strong pri-mary dormancy release and no secondary dormancy induced bycold (low PC2germ) (Fig. 2).We tested for epistatic effects of DOG1 on germination PCs

by dividing the data into 2 subsets of DOG1 haplotypes (D-haplotypes, n = 158, and E-haplotypes, n = 161) and perform-ing a GWA analysis within each haplotype. We found a peak onChr. 5 for PC2germ for the E-haplotype annotated to AT5G65100,an ethylene-insensitive 3 family protein (SI Appendix, Fig. S9A andTable S7). A SNP associated to PC2germ in accessions carrying theE-haplotype falls within a region on Chr. 5 linked to a fitnessquantitative trait loci related to seed coat alterations (67) and seedcoat mucilage production (68). Despite not being significantlyassociated to PC2germ in the D-haplotype accessions, this SNP hasa similar allele frequency among the D- and E-haplotypes (SIAppendix, Table S8). Thus, this association arises not because thecausative variant is absent in 1 haplotypic background, but ratherbecause different SNPs have different effect sizes in differentbackgrounds. The epistatic interaction between the DOG1 E-haplotype and the associated allele in this region of Chr. 5 giveadditional insight into the complex gene interactions involved inthe regulation of germination responses to chilling. Although notmeasured in our study, it is likely that DOG1 expression levels areinvolved in the chilling response as well (12, 53, 57, 59, 62, 63, 69–73). Other mechanisms regulating DOG1 expression might also beinvolved in the chilling germination responses: For example, theantisense transcript asDOG1, which negatively regulates the ex-pression of DOG1 (74, 75).

Life-History Syndromes Resulted from Correlated Selection on MultipleLoci.Accessions with loss-of-function DOG1 D-haplotypes showedstrong primary dormancy release and no secondary dormancyinduced by cold, and also flowered and senesced early. In contrast,individuals with self-binding DOG1 E-haplotypes showed sec-ondary dormancy induced by prolonged cold, and flowered andsenesced later on average (Fig. 2). Besides PC2germ, DOG1 haplo-types explained variation in DTF and DTS (SI Appendix, Fig. S10and Table S6A). Similar D- and E-haplotype germination patternswere observed for geographically restricted samples within Spainand Sweden (SI Appendix, Figs. S11 and S12 and Tables S6 B andC). Pleiotropic effects of DOG1 on flowering and germinationmediated by miR156 and miR172 may be a mechanism behindthese trait correlations (60, 76).Our results also showed that GWAs for DTF were polygenic

(39 SNPs organized in 11 LD blocks) (Fig. 3 and SI Appendix,Table S2) and SNPs tagging DOG1 were not significantly asso-ciated with DTF. DTS showed no associated SNPs above apermutation threshold for significance (SI Appendix, Fig. S5 andTable S2). However, the 5 SNPs associated to PC2germ taggingDOG1 have been previously associated to different flowering-time phenotypes under various conditions (Fig. 3 and SI Ap-pendix, Fig. S13 and Table S2) (42, 43, 64, 77). The lack of as-sociation of DTF with DOG1 in our study could partly be due tothe accessions sampled, as we performed an additional GWAanalysis using only our accessions with subsampled 1001 Ge-nomes Consortium flowering phenotypes (43) and did not findevidence of association of DTF with SNPs tagging DOG1 (SI

0

30

60

90

120

150

180

−0.10 −0.05 0.00 0.05 0.10PC2germ

DTF

A

0

100

200

300

400

−0.10 −0.05 0.00 0.05 0.10PC2germ

DTS

B

D−RYD−SYECCYECSYEFSYunknown

Fig. 2. Correlations between PC2germ and DTF (A) and DTS (B). Colors in-dicate the amino acid sequence at the self-binding domain for differentDOG1 haplotypes.

4 of 9 | www.pnas.org/cgi/doi/10.1073/pnas.1912451117 Martínez-Berdeja et al.

Dow

nloa

ded

by g

uest

on

May

19,

202

0

Appendix, Fig. S14). Our sampling includes mostly Europeanaccessions while the 1001 Genomes phenotypes also includeAsian accessions (78). Moreover, the lack of association couldalso be due to the different vernalization and growth conditionsused in each experiment.We also tested for epistatic effects of DOG1 haplotypes on

flowering and senescence and found that SNPs significantly as-sociated with DTF for the E-haplotype were located at differentpositions from those identified with the whole dataset, and themost significantly associated SNP on Chr. 1 was found closest toAT1G33440, a major facilitator superfamily protein (Fig. 3 andSI Appendix, Fig. S9B and Tables S2, S7, and S8).Taken together, our results suggest that correlations between

seed chilling responses and reproductive timing may result inwinter and summer annual life histories. DOG1 variation iscritical for cold induction of secondary dormancy that deter-mines seasonal germination timing, and may also have pleio-tropic effects on flowering time. However, a number of other lociare important to flowering-time variation, suggesting that themultivariate life-history syndromes we observe may be shaped bycorrelated selection at multiple loci in addition to DOG1.

Arabidopsis Germination Niche Today Is Explained by Current andPast Climate. To understand the evolution of DOG1, we con-structed a gene tree of the first exon of DOG1, which includesthe self-binding domain (Fig. 4A). The earliest DOG1 di-versification represented in extant individuals is that of the an-cestral and widely distributed ECCY haplotype, associated withtraits that promote a winter annual life history. The tree topologyshowed a deep divergence within the ancestral ECCY clade,especially for relict, African, and Swedish accessions and a fewanciently diverged haplotypes in China (Fig. 4A and SI Appendix,Figs. S6 and S7, and Table S9).The rare ECSY haplotype arose within the ECCY clade, fol-

lowed by EFSY, during periods of cooler climate after the lastinterglacial period (Fig. 4A and SI Appendix, Table S9); both ofthese derived haplotypes are geographically restricted to regions

in Iberia with hot, dry summers and are associated with relict andSpanish accessions (Fig. 4 and SI Appendix, Figs. S6, S7, andS15). ECSY and EFSY likely originated in the Iberian refugium(46, 79–81), and may remain endemic in that region because theyrequire hot, dry Mediterranean summers to persist.D-SY and D-RY haplotypes are sister to one another, arising

from an ECCY ancestor (Fig. 4A). Pairwise diversity within allD-clade individuals suggests this split occurred about 365 Kya(ranging from 281 to 450 Kya using differing assumptions aboutgeneration time) (Materials and Methods) during a period ofclimate cooling following an interglacial period (Fig. 4B and SIAppendix, Table S9), possibly during contraction into differentrefugia. However, we do not observe further diversification withinthe extant D-haplotypes until much later, potentially due to cyclesof expansion and contraction out of refugia followed by local ex-tinction. The extant, predominately Asian and east Eurasia D-SYclade diversified 115 Kya (88 to 141 Kya with varying generationtimes) during a cool period of the last interglacial period (Fig. 4Band SI Appendix, Table S9). Pairwise diversity among extant D-RYhaplotypes suggests this west Eurasian and North American cladeexpanded recently 16 Kya (ranging 12 to 19 Kya), during rapidwarming after the last glacial maximum (Fig. 4B and SI Appendix,Table S9), but the gene tree suggests an earlier origin (Fig. 4A).The extant distribution of D-RY in Iberia and Western Europe(SI Appendix, Figs. S6 and S7) suggests that this haplotype mayhave arisen and spread from the Iberian refugium.These observations suggest that changing Pleistocene climate

may have favored the rise and spread of the D-haplotypes,possibly through selection for associated traits promoting springannual or rapid cycling life histories. The loss of cold-inducedsecondary dormancy in D-haplotypes, correlated with shorter lifecycles, may have provided life-history flexibility facilitating per-sistence in changing climates and the invasion of new habitats.The ability to germinate in spring is advantageous in montaneregions (17), such as Central Asia, and rapid cycling life historiesmay be favored in disturbed habitats with ample summer rainfall.Both D-SY and D-RY currently occupy climates with wettersummers than the E-haplotypes (SI Appendix, Fig. S15B). Thewide Eurasian distribution of D-SY haplotypes suggests thattheir expanded germination niche may have facilitated post-glacial expansion and spread of “nonrelict” genotypes out ofcentral Asia into Europe and China (47, 82). The recent ex-pansion of the D-RY clade is consistent with invasion of newruderal habitats made available by the spread of agriculturethrough Europe (43, 82, 83). Our results show that derived loss-of-function D-haplotypes are more common in midlatitudes andEngland, and could have resulted from the nonrelict east–westexpansion following human-mediated habitat disturbance.Relict populations in the Iberian Peninsula and Scandinaviamight have preserved the ancestral DOG1 E-haplotype that pro-motes a winter annual cycle, advantageous in more undisturbedhabitats (81, 82). Moreover, the D-RY haplotype of DOG1 isenriched in the invaded range of North America (SI Appendix,Fig. S6), where A. thaliana has arrived as a human commensalin the last 300 y (84). Although this could be due to a foundereffect, phenological traits associated to this haplotype mightalso have played a role in invading disturbed habitats suitablefor rapid cycling.We caution that selection, geographic sampling, and de-

mography could affect these estimates of age. But, with the ex-ception of a more recent D-RY origin, these clade ages matchthe relative branching pattern on the gene tree. The origin ofDOG1 haplotype variants suggests that the germination niche ofA. thaliana has been shaped by climate cycling throughout thePleistocene, as the species range repeatedly expanded out of andcontracted into glacial refugia (46, 79, 80) and the changingclimate and physical environment filtered which individuals wereable to persist (85). Novel germination strategies, measured via

Fig. 3. Chromosome location (Mb) and association P value from GWAS forthe SNPs on PC1germ, PC2germ, and DTF (permutation-based significancethresholds are shown by dotted horizontal lines). The vertical dashed greyline indicates the location of DOG1 gene.

Martínez-Berdeja et al. PNAS Latest Articles | 5 of 9

EVOLU

TION

Dow

nloa

ded

by g

uest

on

May

19,

202

0

the origin of DOG1 haplotypes, arose at key timepoints in thechanging climate of the Quaternary. Thus, life-history strategiesassociated with DOG1 haplotypes may in fact shape the pop-ulation structure we see today on the landscape.

SNPs Associated with Chilling Responses and DOG1 FunctionalHaplotypes Show a Signature of Climate Adaptation. To under-stand the climatic associations between genetic variants associatedto various phenotypes, we used Gradient Forest, a tree-basedmachine-learning regression approach, to describe nonlinearturnover functions of allele frequencies along environmentalgradients (40, 86, 87). We performed Gradient Forest regres-sion analyses between environmental gradients and LD-prunedindex SNPs from the 1,000 most highly associated SNPs fromour GWA results for PC1germ (23 SNPs), PC2germ (9 SNPs), andthe DOG1 haplotypes. After accounting for spatial autocorre-lation by including Moran’s eigenvector map (MEM) variables,mean temperature of the wettest quarter (Bio8) explained thehighest amount of turnover in allele frequency of SNPs asso-ciated to PC2germ and was among the top 3 predictors forPC1germ as well (Fig. 5 and SI Appendix, Table S10). Addi-tionally, altitude, isothermality (Bio3), and temperature meandiurnal range (Bio2) predicted allele frequencies of both ger-mination PCs (SI Appendix, Fig. S16 and Table S10). Theseenvironmental gradients structure turnover in index SNPs morethan a set of 150 random SNPs (Fig. 5 and SI Appendix, Fig.S16), evidence of local adaptation to climate at these loci. Giventhat the E- and D-haplotypes are very old, our data suggest thatecotypes carrying both of these haplotypes could have arrived atlocations throughout the range, but that environmental filteringand selection likely give rise to the geographic patterns that wesee today. As a predominately selfing annual plant, A. thalianapopulations are structured across the landscape, giving rise togeographic differences in ecologically relevant alleles (88);however, our results support local adaptation to climate, as alleleturnover from index SNPs is more strongly structured by tem-perature and precipitation gradients than alleles from randomSNPs.The response-to-cold PC2germ cumulative importance function

for mean temperature of the wettest quarter (Bio8) showed athreshold turnover in allele frequencies around 14 to 15 °C (Fig.5), and alleles from SNPs associated to DOG1 were among theones with the highest importance values for this function (SIAppendix, Table S11). Accessions with the reference allele, withhigh dormancy release under chilling and no secondary dor-mancy, are found in these southern regions, while accessionswith the alternative DOG1 allele have strong secondary dor-mancy. This temperature threshold divides the winter growingseason in southern regions (<14 to 15 °C) from a warm wetsummer (>14 to 15 °C) growing season in northern latitudes,thus the shape of the allele turnover function suggests that ger-mination traits might have been selected by these climatic gradi-ents. Environmental predictors’ importance varied at explainingthe distribution of different DOG1 haplotypes (SI Appendix, Fig.S17 and Tables S12 and S13). Allele distributions of DOG1 mighthave resulted from its effect on germination responses to chillingand on other life-history traits as well. For example, 14 °C actsboth as a temperature threshold for the induction of strong ma-ternal effects on seed dormancy (11, 22) and for vernalizationdisruption in flowering time (89).

ConclusionsThe seasonal germination niche shapes phenology and life his-tory (1, 2, 5–8), and may be an essential component of adapta-tion to climate (11–13). Natural variation in dormancy andgermination responses to seasonal environmental cues is oftenobserved (21), and the genetic basis of variation in primarydormancy and after-ripening has been well studied in A. thaliana.However, much less is known about the genetic basis of naturalvariation in seed responses to seasonal chilling. Our data reveal anew axis of natural variation in germination and dormancy re-sponses to cold that may drive the expression of winter annual vs.spring annual life histories. Moreover, this variation in response

D−RY

D−SY

ECCY

EFSYECSY

−10

−5

0

5

10

0100200300400500

Age (thousand years ago)

ΔTem

pera

ture

(°C

)

allD

D−RY

D−SY

ECSY

EFSY

B

A

Fig. 4. (A) Bayesian phylogeny of DOG1 haplotypes. Haplotype namesrepresent amino acid sequence at the self-binding domain of DOG1. Tri-angle clade heights represent the number of individuals carrying a re-dundant sequence. The A. lyrata outgroup used to root the phylogeny isnot shown. (B) Ages estimated from pairwise nucleotide divergence ofDOG1 haplotype groups with Pleistocene deuterium isotope records (103).Confidence intervals represent a range of generation times from 1.3 (±0.3)y per generation. Pairwise divergence of D-RY appears older in B as branchlengths are not weighted by number of individuals during tree-building(Materials and Methods). In both plots, points reflect divergence of hap-lotypes or groups of haplotypes. Temperatures are presented as devia-tions from the last 1,000-y average.

6 of 9 | www.pnas.org/cgi/doi/10.1073/pnas.1912451117 Martínez-Berdeja et al.

Dow

nloa

ded

by g

uest

on

May

19,

202

0

to seasonal chilling was correlated with natural variation inflowering time, senescence, and after-ripening to form a range oflife-history syndromes. GWA identified several loci associatedwith natural variation in chilling responses, including a knownfunctional variant of the candidate gene DOG1, which showed ageographic signature of adaptation to present-day climate. Aphylogeny of DOG1 haplotypes revealed ancient divergence ofthese functional variants associated with past periods of climatechange. Thus, natural variation in the A. thaliana germinationniche is shaped by both past and present climate.

Materials and MethodsSeed Bulking. Seeds from 559 A. thaliana fully sequenced accessions from the1001 Genomes Project (ABRC stocks) were stratified in 0.15% agar at 4 °C for7 d. Seeds were sown in soil and allowed to germinate for 10 d; seedlingswere vernalized for 6 wk at 4 °C with a 12/12-h photoperiod and weregrown at 14 °C with a 12/12-h photoperiod in a walk-in growth chamber(Conviron E7/2 Controlled Environment Chamber). Two individuals of eachaccession were grown in the same chamber (a total of 1,118 plants), planted2 wk apart in 2 temporal replicate blocks. Plants were watered twice a weekwith nutrient water until they showed 50 to 60% ripe siliques and seeds wereindividually harvested from each plant when 70 to 80% of the siliques were ripe.Fresh harvested seeds from each plant were immediately used for 2 differentexperiments: 1) Cold stratification germination experiments and 2) dry seedstorage germination experiments to test for seed after-ripening times. Phenol-ogy variables were recorded on the maternal plants including DTF and DTS. DTFvaried from 70 to 180 d while DTS ranged from 151 to 448 d (Fig. 2) (90).

Control and Dark Cold Stratification Germination Experiments. Fresh seedswere stored in a drying box with desiccant in the laboratory for 11 d, thenassigned to 13 experimental treatments. Base germination of imbibed freshseeds (control) was assessed under germination-inducing conditions after7 d at 22 °C and 12/12-h photoperiod (80-μmols m s−1 light). We also exposedseeds to 12 stratification treatments. Seeds for these treatments were im-bibed and stratified in boxes in dark chambers for 4, 8, 11, 15, 22, and 32 d at4 °C (simulating winter temperature cues) or 10 °C (simulating early springand autumn temperature cues) (11, 26). After these different cold treatments,

dark-imbibed stratified seeds were put under the same germination-inducingconditions as the control seeds. Total germination was manually scored fromphotographs (i.e., radicle protrusion) after 7 d. Each experimental replicateconsisted of 25 to 30 fresh seeds scattered on a Petri plate (60 × 15 mm) withblue germination paper (Blue Seed Germination Blotter, Anchor Paper) and 35mL of a 2% PPM solution (Plant Preservative Mixture, Caisson Laboratories).PPM was added to prevent fungal growth during the 4- to 32-d stratificationperiod (91). There were 2 replicates for each accession × treatment combina-tion, for a total of 14,534 Petri plates scored. Germination experiments werecarried out continuously for over a year, as fresh seed from individual plantswere harvested when seeds matured across over 500 d, and not duringa single month period (90).

Dry Seed Storage Germination Experiments. Additional germination experi-ments were done to test for seed after-ripening. Fresh seeds were storedunder dry laboratory conditions and were tested for total germination every6 wk until 75% germination was recorded in 2 consecutive tests, which weconsidered as after-ripened seeds. Some accessions were after-ripened after6 wk while some others showed little evidence of after-ripening even after drystorage in the laboratory for up to 788 d. Germination induction conditionswere the same as for the cold stratification experiment. Seed after-ripeningwas assessed by calculating the number of days to 50%germination by fittinga polynomial regression to the dry-stored seed germination data for eachecotype (DSDS50) (92).

Data Analyses.Germination strategies (PCgerm). We excluded accessions that had misleadinglocation data due to being misidentified according to ref. 93. We ran a PCAon mean percent germination from the 2 plantings for the base and coldtreatment logit-transformed germination data (prcomp function in R). PCAscores and loadings were rescaled to the axes SD.Germination responses and phenological associations. We tested for correlationsbetween the first 2 PCs of germination variables, after-ripening (DSDS50), andphenology variables of the mother plants (i.e., DTF and DTS; Pearson’sproduct-moment correlation, cor.test function in R).GWA on germination and phenology traits. GWA analyses were performed on thescores of the first 2 germination PCs, base germination, DSDS50, DTF, and DTSto identify associated genetic polymorphisms using a univariate linear mixedmodel (LMM) corrected for population structure, implemented in GEMMA(94). Genotypes from sampled individuals (n = 498) were obtained from the1001 Genomes Project (3,483,598 SNPs, minor allele frequency [MAF] 0.01,imputed genotypes) (43). Permutation-based thresholds were calculated foreach trait by running a GWA 100 times with phenotypes permuted overgenotypes and getting the average P value of the top fifth quantile from eachanalysis. We annotated SNPs to the closest TAIR10 gene (distanceToNearestfunction in GenomicRanges) with SNPs >1-kb distance from a gene annotatedas intergenic regions. We also used TAGGIT to annotate the 1,000 most highlyassociated SNPs with respect to seed-specific gene ontology categories (95).GWA results were used to group SNPs into LD blocks. SNPs that were associ-ated at a significance threshold of P ≤ 0.0001 and that were not included inother LD blocks were selected as index SNPs. LD blocks around the index SNPswere defined by all other associated SNPs at a significant threshold of P ≤ 0.01that were in LD with the index SNP (r2 = 0.50) within a physical distance of250 kb (default settings from clump command, PLINK).DOG1 functional haplotypes dated phylogeny.A 3-bp indel known to affect DOG1function (53) is not genotyped in the available 1001 Genomes Project vcffiles. In order to characterize our accessions at this known functional site, wemapped whole-genome resequencing reads from 1,135 A. thaliana acces-sions from the 1001 Genomes Consortium (43), 64 African A. thaliana ac-cessions (45), and 118 Chinese A. thaliana accessions (46) to the primarycDNA of DOG1 (AT5G45830.1) using bwa-mem v0.7.12-r1039 (96). Weretained mapped reads and their pair, and assembled these using phrapv1.090518 (http://www.phrap.org). We aligned assemblies to exons of DOG1using MAFFT v7.27 (97), and extracted regions from the haplotype at aminoacid positions 13 to 16, as defined by ref. 53. We filtered minor alleles withfewer than 2 individuals, which eliminated 3 rare haplotypes: C-CY, D-CY,and EYSY. Due to alignment and assembly issues, we recovered sequence for972 individuals. DOG1 local reassemblies of the 393 bp of exon 1 werepruned for redundant sequences, resulting in 74 unique individual se-quences (98, 99) used to generate a Bayesian likelihood phylogeny in BEAST2.5.2 with a strict clock and constant-sized coalescent prior, using an Ara-bidopsis lyrata and Capsella rubella DOG1 sequence as outgroups. We ranBEAST for 10 million generations, confirmed Markov chain Monte Carloconvergence by eye, discarded 10% of trees as burn-in, and summarizedtrees with a maximum clade credibility tree. Attempts to run BEAST including

0.00

0.01

0.02

0.03

0 5 10 15 20mean temp. wettest quarter (°C)

cum

ulat

ive

impo

rtanc

e

PC1germPC2germ

Fig. 5. Average cumulative importance functions of mean temperature ofthe wettest quarter for the index SNPs of the top 100 SNPs with the highestassociation with the first 2 PCs of germination for 752 accessions. Genomiccontrols are shown in grey, representing the 100 cumulative functions of 150random SNPs with MAF > 0.01.

Martínez-Berdeja et al. PNAS Latest Articles | 7 of 9

EVOLU

TION

Dow

nloa

ded

by g

uest

on

May

19,

202

0

all 972 tips, many of which were redundant sequences, failed to converge.Admixture group assignments from the 1001 Genomes Consortium (43) wereused to display the geographic distribution of the haplotypes on the tree, with“Africa” used for samples sourced from Durvasula et al. (45) and “China” usedfor samples sourced from Zou et al. (46). To estimate the timing of origin ofeach DOG1 haplotype, we calculated average pairwise genetic diversity withinthe exon 1 sequence of each haplotype, using pegas (100), and converted toyears using a mutation rate of 7.1e-9 mutations per generation (101) (T = π/2u,where T is time in generations, π is average pairwise diversity, and u is the per-generation mutation rate). As differences in life history can generate varia-tion in generation times, we must make assumptions about generation timeto directly relate generations to absolute time. Although A. thaliana is anannual plant, models suggest all environments except the southernmostSpanish population have generation times longer than 1 y (up to 4 y pergeneration) (7), and seed bank dynamics in northern Scandinavian pop-ulations estimate 1.3 y per generation (102). These differences in life historycould be directly impacted by germination behavior. We use the experi-mental value of 1.3 (±0.3) years per generation (102) for all samples to es-timate absolute time, but caution that relaxing this assumption can give riseto differing relationships to climatic events. Similarly, differences betweenthe mutation rate and substitution rate (36) would affect the relationshipto absolute climatic events. We related these dates to deuterium recordsfrom Jouzel et al. (103) in Fig. 4B. Scripts and files are available at https://github.com/mcstitzer/martinez-berdeja_dog1.DOG1 functional haplotypes associations with germination and phenology.We usedan LMM with population structure as a covariate (relmatLmer function inlme4qtl package in R) to analyze PC1germ, PC2germ, DTF, and DTS differencesamong of DOG1 haplotypes (n = 387). Additionally, we analyzed regionalphenotypic differences among DOG1 haplotypes from Spanish (n = 75) andSwedish (n = 125) accessions.Genomic environmental associations. We ran Gradient Forest (gradientForest Rpackage) to find the environmental variables that best explained spatialvariation in SNP alleles identified by GWA.We selected candidate SNPs to runGradient Forest by LD pruning the 100 most highly associated SNPs from eachGWA output (40, 86). We included genotypes from n = 752 accessions ex-cluding those outside longitude −15 to 90 and randomly including 1 accession

from the ones with identical location data. We used altitude, Bioclim 1 to 19climate variables from the location of origin of each accession (WorldClim)(104), and included 23 MEM variables to account for spatial autocorrela-tion (40). Determining the number of MEMs necessary to account forpopulation structure is not clear. We used a similar number of MEMs as theenvironmental variables used as environmental predictors, which corre-sponded to 5% of the MEMs calculated using adespatial R package fol-lowing Jansen et al. (87). The top-ranked environmental variables wererobust to inclusion of different numbers of MEMs. To assess whether theobserved environmental associations were more extreme than thoseexpected due to neutral processes, we tested a set of a random SNPs,resampling 100 times. These genomic controls were obtained by runningGradient Forest on 150 random SNPs, with a MAF > 0.01 and a similar LD(r2 = 0.5 in a 250 k window that initiated every 10 SNPs using Plink). We alsoran Gradient Forest on the different DOG1 haplotypes to find the environ-mental variables that best explained haplotype distribution along environ-mental gradients (GradientForest R package). For this analysis we used theaccessions from 1001 Genomes for which we assembled DOG1 haplotypesand selected a set of n = 516 accessions, excluding the ones located outsidelongitude −15 to 90 and including a single representative from repeatedlocation data.Data availability. Germination, phenology and DOG1 haplotype data areavailable on Dryad repository and AraPheno. DOG1 haplotype data andcode are also available on GitHub (https://github.com/mcstitzer/martinez-berdeja_dog1).

ACKNOWLEDGMENTS. We thank Kent Bradford, Moises Exposito-Alonso,Arthur Korte, Jeff Ross-Ibarra, Liana Burghardt, Emily Josephs, and MeganBontrager for valuable advice and comments on the manuscript; and MireilleCaton-Darby, Holly Addington, Danielle Ethington, Felicia Wong, JoshLeung, Bryan González, Helena Bayat, Lydia Eldridge, and Jasneek Attwalfor growing the accessions and carrying out the germination experiments.This work was supported by a Conacyt Postdoctoral fellowship (to A.M.-B.);National Science Foundation Grants DEB-1447203 and DEB-1754102 (to J.S.);and the University of California, Davis.

1. K. Donohue, Germination timing influences natural selection on life-history char-acters in Arabidopsis thaliana. Ecology 83, 1006–1016 (2002).

2. K. Donohue, Niche construction through phenological plasticity: Life history dy-namics and ecological consequences. New Phytol. 166, 83–92 (2005).

3. L. F. Galloway, Parental environmental effects on life history in the herbaceous plant

Campanula americana. Ecology 82, 2781–2789 (2001).4. A. M. Wilczek et al., Effects of genetic perturbation on seasonal life history plasticity.

Science 323, 930–934 (2009).5. X. Huang et al., The earliest stages of adaptation in an experimental plant population:

Strong selection on QTLS for seed dormancy. Mol. Ecol. 19, 1335–1351 (2010).6. G. C. Chiang et al., Pleiotropy in the wild: The dormancy gene DOG1 exerts cascading

control on life cycles. Evolution 67, 883–893 (2013).7. L. T. Burghardt, C. J. Metcalf, A. M. Wilczek, J. Schmitt, K. Donohue, Modeling the

influence of genetic and environmental variation on the expression of plant lifecycles across landscapes. Am. Nat. 185, 212–227 (2015).

8. L. T. Burghardt, B. R. Edwards, K. Donohue, Multiple paths to similar germinationbehavior in Arabidopsis thaliana. New Phytol. 209, 1301–1312 (2016).

9. J. R. Gremer, C. J. Wilcox, A. Chiono, E. Suglia, J. Schmitt, Germination timing andchilling exposure create contingency in life history and influence fitness in the nativewildflower Streptanthus tortuosus. J. Ecol. 108, 239–255 (2020).

10. L. F. Galloway, J. R. Etterson, Transgenerational plasticity is adaptive in the wild.Science 318, 1134–1136 (2007).

11. A. Montesinos-Navarro, F. X. Picó, S. J. Tonsor, Clinal variation in seed traits influ-encing life cycle timing in Arabidopsis thaliana. Evolution 66, 3417–3431 (2012).

12. D. S. Vidigal et al., Altitudinal and climatic associations of seed dormancy andflowering traits evidence adaptation of annual life cycle timing in Arabidopsisthaliana. Plant Cell Environ. 39, 1737–1748 (2016).

13. F. M. Postma, J. Ågren, Early life stages contribute strongly to local adaptation inArabidopsis thaliana. Proc. Natl. Acad. Sci. U.S.A. 113, 7590–7595 (2016).

14. S. Kalisz, Variable selection on the timing of germination in Collinsia verna(Scrophulariaceae). Evolution 40, 479–491 (1986).

15. K. Donohue et al., The evolutionary ecology of seed germination of Arabidopsisthaliana: Variable natural selection on germination timing. Evolution 59, 758–770(2005).

16. T. M. Korves et al., Fitness effects associated with the major flowering time geneFRIGIDA in Arabidopsis thaliana in the field. Am. Nat. 169, E141–E157 (2007).

17. F. X. Picó, Demographic fate of Arabidopsis thaliana cohorts of autumn- and spring-germinated plants along an altitudinal gradient. J. Ecol. 100, 1009–1018 (2012).

18. L. T. Burghardt, C. J. Metcalf, K. Donohue, A cline in seed dormancy helps conservethe environment experienced during reproduction across the range of Arabidopsisthaliana. Am. J. Bot. 103, 47–59 (2016).

19. M. Debieu et al., Co-variation between seed dormancy, growth rate and floweringtime changes with latitude in Arabidopsis thaliana. PLoS One 8, e61075 (2013).

20. A. Marcer et al., Temperature fine-tunes Mediterranean Arabidopsis thaliana life-cycle phenology geographically. Plant Biol (Stuttg) 20 (suppl. 1), 148–156 (2018).

21. C. C. Baskin, J. M. Baskin, Seeds. Ecology, Biogeography, and Evolution of Dormancyand Germination (San Diego Academic Press, ed. 2, 2014).

22. S. Penfield, V. Springthorpe, Understanding chilling responses in Arabidopsis seedsand their contribution to life history. Philos. Trans. R. Soc. Lond. B Biol. Sci. 367, 291–297 (2012).

23. S. Footitt, Z. Huang, H. A. Clay, A. Mead, W. E. Finch-Savage, Temperature, light andnitrate sensing coordinate Arabidopsis seed dormancy cycling, resulting in winterand summer annual phenotypes. Plant J. 74, 1003–1015 (2013).

24. W. E. Finch-Savage, S. Footitt, Seed dormancy cycling and the regulation of dor-mancy mechanisms to time germination in variable field environments. J. Exp. Bot.68, 843–856 (2017).

25. J. R. Gremer, S. Kimball, D. L. Venable, Within-and among-year germination inSonoran Desert winter annuals: Bet hedging and predictive germination in a vari-able environment. Ecol. Lett. 19, 1209–1218 (2016).

26. G. A. Auge et al., Secondary dormancy dynamics depends on primary dormancystatus in Arabidopsis thaliana. Seed Sci. Res. 25, 230–246 (2015).

27. L. Bentsink, J. Jowett, C. J. Hanhart, M. Koornneef, Cloning of DOG1, a quantitativetrait locus controlling seed dormancy in Arabidopsis. Proc. Natl. Acad. Sci. U.S.A. 103,17042–17047 (2006).

28. L. Bentsink, M. Koornneef, Seed dormancy and germination. Arabidopsis Book 6,e0119 (2008).

29. S. Kendall, S. Penfield, Maternal and zygotic temperature signalling in the control ofseed dormancy and germination. Seed Sci. Res. 22, S23–S29 (2012).

30. M. Chen et al., Maternal temperature history activates Flowering Locus T in fruits tocontrol progeny dormancy according to time of year. Proc. Natl. Acad. Sci. U.S.A.111, 18787–18792 (2014).

31. H. He et al., Interaction between parental environment and genotype affects plantand seed performance in Arabidopsis. J. Exp. Bot. 65, 6603–6615 (2014).

32. S. Footitt, I. Douterelo-Soler, H. Clay, W. E. Finch-Savage, Dormancy cycling in Ara-bidopsis seeds is controlled by seasonally distinct hormone-signaling pathways. Proc.Natl. Acad. Sci. U.S.A. 108, 20236–20241 (2011).

33. S. Footitt, H. A. Clay, K. Dent, W. E. Finch-Savage, Environment sensing in spring-dispersed seeds of a winter annual Arabidopsis influences the regulation of dormancyto align germination potential with seasonal changes. New Phytol. 202, 929–939 (2014).

34. W. E. Finch-Savage, G. Leubner-Metzger, Seed dormancy and the control of germi-nation. New Phytol. 171, 501–523 (2006).

35. J. D. Bewley, K. Bradford, H. Hilhorst, H. Nonogaki, Seeds: Physiology of Develop-ment, Germination and Dormancy (Springer, ed. 3, 2013).

8 of 9 | www.pnas.org/cgi/doi/10.1073/pnas.1912451117 Martínez-Berdeja et al.

Dow

nloa

ded

by g

uest

on

May

19,

202

0

36. M. Exposito-Alonso et al., The rate and potential relevance of new mutations in acolonizing plant lineage. PLoS Genet. 14, e1007155 (2018).

37. M. Takou et al., Linking genes with ecological strategies in Arabidopsis thaliana. J.Exp. Bot. 70, 1141–1151 (2019).

38. J. R. Etterson, R. G. Shaw, Constraint to adaptive evolution in response to globalwarming. Science 294, 151–154 (2001).

39. A. Fournier-Level et al., A map of local adaptation in Arabidopsis thaliana. Science334, 86–89 (2011).

40. M. C. Fitzpatrick, S. R. Keller, Ecological genomics meets community-level modellingof biodiversity: Mapping the genomic landscape of current and future environ-mental adaptation. Ecol. Lett. 18, 1–16 (2015).

41. M. Exposito-Alonso et al., Genomic basis and evolutionary potential for extremedrought adaptation in Arabidopsis thaliana. Nat. Ecol. Evol. 2, 352–358 (2018).

42. S. Atwell et al., Genome-wide association study of 107 phenotypes in Arabidopsisthaliana inbred lines. Nature 465, 627–631 (2010).

43. 1001 Genomes Consortium, 1,135 Genomes reveal the global pattern of poly-morphism in Arabidopsis thaliana. Cell 166, 481–491 (2016).

44. J. B. Beck, H. Schmuths, B. A. Schaal, Native range genetic variation in Arabidopsisthaliana is strongly geographically structured and reflects Pleistocene glacial dy-namics. Mol. Ecol. 17, 902–915 (2008).

45. A. Durvasula et al., African genomes illuminate the early history and transition toselfing in Arabidopsis thaliana. Proc. Natl. Acad. Sci. U.S.A. 114, 5213–5218 (2017).

46. Y. P. Zou et al., Adaptation of Arabidopsis thaliana to the Yangtze River basin.Genome Biol. 18, 239 (2017).

47. C. W. Hsu, C. Y. Lo, C. R. Lee, On the postglacial spread of human commensal Ara-bidopsis thaliana: Journey to the East. New Phytol. 222, 1447–1457 (2019).

48. J. R. Stinchcombe et al., A latitudinal cline in flowering time in Arabidopsis thalianamodulated by the flowering time gene FRIGIDA. Proc. Natl. Acad. Sci. U.S.A. 101,4712–4717 (2004).

49. F. Bouché, G. Lobet, P. Tocquin, C. Périlleux, FLOR-ID: An interactive database offlowering-time gene networks in Arabidopsis thaliana. Nucleic Acids Res. 44, D1167–D1171 (2016).

50. E. Kerdaffrec et al., Multiple alleles at a single locus control seed dormancy inSwedish Arabidopsis. eLife 5, e22502 (2016).

51. I. Kronholm, F. X. Picó, C. Alonso-Blanco, J. Goudet, J. de Meaux, Genetic basis ofadaptation in Arabidopsis thaliana: Local adaptation at the seed dormancy QTLDOG1. Evolution 66, 2287–2302 (2012).

52. F. M. Postma, J. Ågren, Maternal environment affects the genetic basis of seeddormancy in Arabidopsis thaliana. Mol. Ecol. 24, 785–797 (2015).

53. K. Nakabayashi, M. Bartsch, J. Ding, W. J. Soppe, Seed dormancy in Arabidopsis re-quires self-binding ability of DOG1 protein and the presence of multiple isoformsgenerated by alternative splicing. PLoS Genet. 11, e1005737 (2015).

54. D. Tabas-Madrid et al., Genome-wide signatures of flowering adaptation to climatetemperature: Regional analyses in a highly diverse native range of Arabidopsisthaliana. Plant Cell Environ. 41, 1806–1820 (2018).

55. B. R. Ni, K. J. Bradford, Germination and dormancy of abscisic acid- and gibberellin-deficient mutant tomato (Lycopersicon esculentum) seeds. Plant Physiol. 101, 607–617 (1993).

56. G. Leubner-Metzger, Functions and regulation of β-1,3-glucanases during seedgermination, dormancy release and after-ripening. Seed Sci. Res. 13, 17–34 (2003).

57. A. Voegele, A. Linkies, K. Müller, G. Leubner-Metzger, Members of the gibberellinreceptor gene family GID1 (GIBBERELLIN INSENSITIVE DWARF1) play distinct rolesduring Lepidium sativum and Arabidopsis thaliana seed germination. J. Exp. Bot. 62,5131–5147 (2011).

58. G. Née et al., DELAY OF GERMINATION1 requires PP2C phosphatases of the ABAsignalling pathway to control seed dormancy. Nat. Commun. 8, 72 (2017).

59. K. Nakabayashi et al., The time required for dormancy release in Arabidopsis isdetermined by DELAY OF GERMINATION1 protein levels in freshly harvested seeds.Plant Cell 24, 2826–2838 (2012).

60. H. Huo, S. Wei, K. J. Bradford, DELAY OF GERMINATION1 (DOG1) regulates bothseed dormancy and flowering time through microRNA pathways. Proc. Natl. Acad.Sci. U.S.A. 113, E2199–E2206 (2016).

61. A. Linkies et al., Ethylene interacts with abscisic acid to regulate endosperm ruptureduring germination: A comparative approach using Lepidium sativum and Arabi-dopsis thaliana. Plant Cell 21, 3803–3822 (2009).

62. K. Graeber et al., DELAY OF GERMINATION 1 mediates a conserved coat-dormancymechanism for the temperature- and gibberellin-dependent control of seed ger-mination. Proc. Natl. Acad. Sci. U.S.A. 111, E3571–E3580 (2014).

63. G. C. Chiang et al., DOG1 expression is predicted by the seed-maturation environ-ment and contributes to geographical variation in germination in Arabidopsisthaliana. Mol. Ecol. 20, 3336–3349 (2011).

64. Y. Li, Y. Huang, J. Bergelson, M. Nordborg, J. O. Borevitz, Association mapping oflocal climate-sensitive quantitative trait loci in Arabidopsis thaliana. Proc. Natl. Acad.Sci. U.S.A. 107, 21199–21204 (2010).

65. M. A. Taylor et al., Interacting effects of genetic variation for seed dormancy andflowering time on phenology, life history, and fitness of experimental Arabidopsisthaliana populations over multiple generations in the field. New Phytol. 216, 291–302 (2017).

66. J. R. Gremer, D. L. Venable, Bet hedging in desert winter annual plants: Optimalgermination strategies in a variable environment. Ecol. Lett. 17, 380–387 (2014).

67. N. Price et al., Combining population genomics and fitness QTLs to identify thegenetics of local adaptation in Arabidopsis thaliana. Proc. Natl. Acad. Sci. U.S.A. 115,5028–5033 (2018).

68. J. F. Golz et al., Layers of regulation—Insights into the role of transcription factors con-trolling mucilage production in the Arabidopsis seed coat. Plant Sci. 272, 179–192 (2018).

69. H. Nonogaki, O. H. Gee, K. J. Bradford, A germination-specific endo-β-mannanasegene is expressed in the micropylar endosperm cap of tomato seeds. Plant Physiol.123, 1235–1246 (2000).

70. F. Chen, K. J. Bradford, Expression of an expansin is associated with endospermweakening during tomato seed germination. Plant Physiol. 124, 1265–1274 (2000).

71. W. Chen et al., Expression profile matrix of Arabidopsis transcription factor genessuggests their putative functions in response to environmental stresses. Plant Cell 14,559–574 (2002).

72. Y. Yamauchi et al., Activation of gibberellin biosynthesis and response pathways bylow temperature during imbibition of Arabidopsis thaliana seeds. Plant Cell 16, 367–378 (2004).

73. S. L. Kendall et al., Induction of dormancy in Arabidopsis summer annuals requiresparallel regulation of DOG1 and hormone metabolism by low temperature and CBFtranscription factors. Plant Cell 23, 2568–2580 (2011).

74. H. Fedak et al., Control of seed dormancy in Arabidopsis by a cis-acting noncodingantisense transcript. Proc. Natl. Acad. Sci. U.S.A. 113, E7846–E7855 (2016).

75. R. Yatusevich et al., Antisense transcription represses Arabidopsis seed dormancyQTL DOG1 to regulate drought tolerance. EMBO Rep. 18, 2186–2196 (2017).

76. G. A. Auge, S. Penfield, K. Donohue, Pleiotropy in developmental regulation by flowering-pathway genes: Is it an evolutionary constraint? New Phytol. 224, 55–70 (2019).

77. M. Togninalli et al., The AraGWAS catalog: A curated and standardized Arabidopsisthaliana GWAS catalog. Nucleic Acids Res. 46, D1150–D1156 (2018).

78. Y. Zan, Ö. Carlborg, A polygenic genetic architecture of flowering time in theworldwide Arabidopsis thaliana population. Mol. Biol. Evol. 36, 141–154 (2019).

79. H. P. Comes, J. W. Kadereit, The effect of Quaternary climatic changes on plantdistribution and evolution. Trends Plant Sci. 3, 432–438 (1998).

80. T. F. Sharbel, B. Haubold, T. Mitchell-Olds, Genetic isolation by distance in Arabi-dopsis thaliana: Biogeography and postglacial colonization of Europe. Mol. Ecol. 9,2109–2118 (2000).

81. B. Toledo, A. Marcer, B. Méndez-Vigo, C. Alonso-Blanco, F. X. Picó, An ecologicalhistory of the relict genetic lineage of Arabidopsis thaliana. Environ. Exp. Bot.,10.1016/j.envexpbot.2019.103800 (2019).

82. C. R. Lee et al., On the post-glacial spread of human commensal Arabidopsis thali-ana. Nat. Commun. 8, 14458 (2017).

83. O. François, M. G. Blum, M. Jakobsson, N. A. Rosenberg, Demographic history ofeuropean populations of Arabidopsis thaliana. PLoS Genet. 4, e1000075 (2008).

84. J. Hagmann et al., Century-scale methylome stability in a recently diverged Arabi-dopsis thaliana lineage. PLoS Genet. 11, e1004920 (2015).

85. G. M. Hewitt, Some genetic consequences of ice ages, and their role in divergenceand speciation. Biol. J. Linn. Soc. Lond. 58, 247–276 (1996).

86. R. A. Bay et al., Genomic signals of selection predict climate-driven population de-clines in a migratory bird. Science 359, 83–86 (2018).

87. A. Jansen van Rensburg, M. Cortazar-Chinarro, A. Laurila, J. Van Buskirk, Adaptivegenomic variation associated with environmental gradients along a latitudinal clinein Rana temporaria. BioRxiv https://doi.org/10.1101/427872 (27 September 2018).

88. C. Toomajian et al., A nonparametric test reveals selection for rapid flowering in theArabidopsis genome. PLoS Biol. 4, e137 (2006).

89. S. Duncan et al., Seasonal shift in timing of vernalization as an adaptation to ex-treme winter. eLife 4, e06620 (2015).

90. A. Martínez-Berdeja, J. Schmitt, arabidopsis_germ_phen_chilling_DATA.xls. Dryad.https://doi.org/10.25338/B8VS4P. Deposited 10 December 2019.

91. H. Huo et al., Rapid identification of lettuce seed germination mutants by bulkedsegregant analysis and whole genome sequencing. Plant J. 88, 345–360 (2016).

92. C. Alonso-Blanco, L. Bentsink, C. J. Hanhart, H. Blankestijn-de Vries, M. Koornneef,Analysis of natural allelic variation at seed dormancy loci of Arabidopsis thaliana.Genetics 164, 711–729 (2003).

93. A. E. Anastasio et al., Source verification of mis-identified Arabidopsis thaliana ac-cessions. Plant J. 67, 554–566 (2011).

94. X. Zhou, P. Carbonetto, M. Stephens, Polygenic modeling with bayesian sparse linearmixed models. PLoS Genet. 9, e1003264 (2013).

95. E. Carrera et al., Gene expression profiling reveals defined functions of the ATP-binding cassette transporter COMATOSE late in phase II of germination. PlantPhysiol. 143, 1669–1679 (2007).

96. H. Li, Aligning sequences reads, clone sequences and assembly contigs with BWA-MEM. arXiv:1303.3997 (26 May 2013).

97. K. Katoh, D. M. Standley, MAFFT multiple sequence alignment software version 7:Improvements in performance and usability. Mol. Biol. Evol. 30, 772–780 (2013).

98. M. C. Stitzer, dog1.may22.uniq.txt. Dryad. https://doi.org/10.25338/B8VS4P. De-posited 10 December 2019.

99. M. C. Stitzer, Scripts to generate DOG1 haplotypes and understand their evolutionfor Martinez-Berdeja et al., 2019. GitHub. https://github.com/mcstitzer/martinez-berdeja_dog1. Deposited 10 December 2019.

100. E. Paradis, pegas: An R package for population genetics with an integrated-modularapproach. Bioinformatics 26, 419–420 (2010).

101. S. Ossowski et al., The rate and molecular spectrum of spontaneous mutations inArabidopsis thaliana. Science 327, 92–94 (2010).

102. M. Falahati-Anbaran, S. Lundemo, H. K. Stenøien, Seed dispersal in time can coun-teract the effect of gene flow between natural populations of Arabidopsis thaliana.New Phytol. 202, 1043–1054 (2014).

103. J. Jouzel et al., Orbital and millennial Antarctic climate variability over the past800,000 years. Science 317, 793–796 (2007).

104. R. J. Hijmans, S. E. Cameron, J. L. Parra, P. G. Jones, A. Jarvis, Very high resolution in-terpolated climate surfaces for global land areas. Int. J. Climatol. 25, 1965–1978 (2005).

Martínez-Berdeja et al. PNAS Latest Articles | 9 of 9

EVOLU

TION

Dow

nloa

ded

by g

uest

on

May

19,

202

0